Need help fixing recurring black screens

[cubtekki]

Hey everyone this is my first post here but this place is great, I've gotten help by reading these forums many times before.

The problem I've been having with my computer started three or so months back, it seemed to go away for weeks at a time, but now it's really bugging me since it's preventing me from playing any games and the instability can cause work loss.

I get some random and some crashes where I can reliably expect them both when stressed and when not stressed. I've reformatted windows, tried different drivers, used different memory slots and speeds, and unplugged all unnecessary peripherals (only mouse,keyboard,dp-video,power). Took a day and ran memtest86 on all modules one at a time in slot a2 at 3200mhz with no errors on any module. I repeated the test in dual channel with slots a2 and b2 with no errors as well. The problems start when testing any combination involving slots a1 and b1 with errors showing up right away at 2800mhz and higher. The only other parts I can potentially test are another GPU (R9 270) and a different PSU (500W) although since those parts are so different I'm not sure testing with them will help.

The problems manifested in either random blue screens (PAGE_FAULT_IN_NONPAGED_AREA, SYSTEM_SERVICE_EXCEPTION, and nvidia driver error one time) or black screens where video just cuts out but the computer is still on with audio playback, hardware lights and fans spinning (gpu spins full speed).

I logged everything using hwinfo64 when I expected a crash during a stress test (heaven benchmark reliably causes black screen < 2min) and I saw that the pci-e slot seems to stop working or something (showing 0x pci-e speed). I already RMA'd my first GPU due to artifacting thinking it was failing but there was no improvement, in fact I think it got worse. The memory failures even though each module is individually good combined with the GPU slot issues makes me think the problem is the motherboard, but a few other possibilities are bouncing around and just wondering what kind of ideas you all might have. Any help figuring out what it is that I should send in next is greatly appreciated.

 

system info:

Asrock B450M Pro4

AMD Ryzen 5 2600 @ stock speed

Team Group 32GB (4 x 8GB) DDR4 3200mhz, B-Die (tested @ 2400mhz w/crashing also)

EVGA GTX 1080 FTW 8GB

Windows 10 Pro

Samsung 970 Pro M.2 (os ssd)

EVGA 750W P2

[cubtekki]

Just want to update everyone on where I'm at in fixing this in case anyone else has been having similar issues.

==================

-TLDR-

I'd appreciate some feedback on whether the variations in voltage I'm seeing in the bios/hwinfo are anything to worry about. The reading were [12V:12.249-12.302 5V:5.064 3V:3.296-3.312] (in bios/A-tuning) and [12V:11.088-11.141 5V:3.12 3V:3.472] (hwinfo). Maybe it's a sign of a failing PSU or enough to trigger voltage protections on a 1080? Also is the discrepancy I observed in hwinfo more than likely a result of a software error, if so what else do you guys recommend for monitoring voltage? Thanks!

==================
I found the GPU to be defective, and the instability is most likely a result of using more memory at a higher speed than the motherboard was intended for (4x8Gb @3200). Before my replacement 1080 arrived, I tested a R9 270 without it showing any issues, and then I tested the 1080 in another system with the same results I was getting with it in mine -- black screen/fans 100% after running benchmarks for a few minutes. It seems like I'm not the only one to run into this issue either according to this EVGA thread,
https://forums.evga.com/GTX-1080-FTW-Black-screen-amp-fans-spin-up-to-100-m2530081.aspx
After getting the replacement I haven't had any black screen or crashes during games so I'm considering that issue closed. I still can't run all 32G of memory at the speeds I want (3200Mhz) but that will have to wait for a better motherboard when upgrading to Ryzen 3xxx.

 

The only thing that still concerns me, is when talking with EVGA support, I was asked to go into my bios and have a look at my voltage readings on the 12v rail as well as the 5v and 3v to see if my PSU is the issue. They were [12V:12.249-12.302 5V:5.064 3V:3.296-3.312]. The support rep said this was too inconsistent, and that it probably fluctuates even more at load triggering voltage protection in the card which would cause loss of video signal from the GPU. This seemed like a hasty conclusion and I suspect they tried blaming the PSU since they didn't know I'm using an in-warranty EVGA PSU as well. After telling them the black screens happens in multiple systems they conceded something is probably wrong with the card; however they suggested I send the PSU in as well. I get the hunch that they are simply covering their ass after deflecting blame to my PSU initially. I wasn't completely sure if my voltages were too inconsistent so I resolved to find out for sure if the little of variations are magnified under load.

I logged all data over multiple runs with hwinfo during stress tests and at idle before and after. I found the voltage readings to be really low within the program [12V:11.088-11.141 5V:3.12 3V:3.472]. The readings didn't fluctuate under load, they were just weirdly low the entire time. I checked another monitoring source, "A-Tuning" for my ASRock mobo, and that gave me readings that were healthy and line up exactly with what I was seeing in the bios, so that was a little strange. I then took readings with a multi-meter using a spare 12V cable and observed 12.2V consistently under both load and idle. So it seems my concerns are resolved unless I should take the hwinfo readings as gospel. So would you all send in your PSU for RMA just to get a re-certified one (ridiculous), pay for shipping etc...? Just want to get some feedback, maybe it's a little ocd but I just want to be sure. Thanks!

Edited April 30, 2019 by cubtekki
improved clarity of the post

[cubtekki]

BUMP

Just hoping to get some feedback on this.